Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 50

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x79867f0>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 50

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7b7bc50>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.1.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    inputs_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name='inputs_real')
    inputs_z = tf.placeholder(tf.float32, (None, z_dim), name='inputs_z')
    learning_rate = tf.placeholder(tf.float32)

    return inputs_real, inputs_z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param image: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    with tf.variable_scope('discriminator', reuse=reuse):
        filter_size = 4
        alpha = 0.2
        # 28x28x3 to 56*56*64
        x1 = tf.layers.conv2d(images, 64, filter_size, strides=2, padding='same')
        relu1 = tf.maximum(alpha * x1, x1)
        
        # 56*56*64 to 112*112*128
        x2 = tf.layers.conv2d(relu1, 128, filter_size, strides=2, padding='same')
        bn1 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn1, bn1)
        
        # 112*112*128 to 224*224*256
        x3 = tf.layers.conv2d(relu2, 256, filter_size, strides=2, padding='same')
        bn2 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn2, bn2)
        
        # Flatten
        flat = tf.reshape(relu2, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)

    return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    with tf.variable_scope('generator', reuse=not is_train):
        alpha = 0.2
        
        # First fully connected layer
        # to 2*2*512
        x1 = tf.layers.dense(z, 7*7*512)
        x1 = tf.reshape(x1, (-1, 7, 7, 512))
        bn1 = tf.layers.batch_normalization(x1, training=is_train)
        relu1 = tf.maximum(alpha * bn1, bn1)
        
        # 2*2*512 to 4*4*512
        x2 = tf.layers.conv2d_transpose(relu1, filters=512, kernel_size=4, strides=2, padding='same')
        bn2 = tf.layers.batch_normalization(x2, training=is_train)
        relu2 = tf.maximum(alpha * bn2, bn2)
        
        # 4*4*512 to 16*16*256
        #x3 = tf.layers.conv2d_transpose(relu2, filters=128, kernel_size=4, strides=2, padding='same')
        #bn3 = tf.layers.batch_normalization(x3, training=is_train)
        #relu3 = tf.maximum(alpha * bn3, bn3)
                
        # 28*28*128 to 56*56*5
       # x4 = tf.layers.conv2d_transpose(relu3, 64, 4, strides=2, padding='same')
       # bn4 = tf.layers.batch_normalization(x4, training=is_train)
       # relu4 = tf.maximum(alpha * bn4, bn4)
        
        # 56*56*5 to 28*28*3
        logits = tf.layers.conv2d_transpose(relu2, filters=out_channel_dim, kernel_size=4, strides=2, padding='same')
        
        out = tf.tanh(logits)
        
        #print(tf.shape(out))
    
        return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    
    g_model = generator(input_z, out_channel_dim)    
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)
    
    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real)))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake

    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    variables = tf.trainable_variables()
    d_vars = [var for var in variables if var.name.startswith('discriminator')]
    g_vars = [var for var in variables if var.name.startswith('generator')]
    
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)

    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    
    steps = 0
    print_every = 10
    show_every = 100
    #saver = tf.train.Saver()    
    
    image_channels = 3 if 'RGB' == data_image_mode else 1
    inputs_real, inputs_z, _ = model_inputs(data_shape[1], data_shape[2], image_channels, z_dim)
    
    d_loss, g_loss = model_loss(inputs_real, inputs_z, image_channels)
    
    d_train_opt, g_train_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
        
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                
                steps += 1
                
                # Sample random noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))

                # Review
                # The output from the generator is bound between -1 and 1 as we use tanh as the activation function 
                # in the output layer. The input batch images are in the range -0.5 to 0.5. 
                # You must scale the input images to match the values as in generated image
                batch_images = batch_images * 2
                
                # Run optimizers
                _ = sess.run(d_train_opt, feed_dict={inputs_real: batch_images, inputs_z: batch_z})
                _ = sess.run(g_train_opt, feed_dict={inputs_z: batch_z, inputs_real: batch_images})

                if steps % print_every == 0:
                    # At the end of each epoch, get the losses and print them out
                    train_loss_d = d_loss.eval({inputs_z: batch_z, inputs_real: batch_images})
                    train_loss_g = g_loss.eval({inputs_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epochs),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))

                if steps % show_every == 0:
                    _ = show_generator_output(sess, 10, inputs_z, image_channels, data_image_mode)
          
        #saver.save(sess, './checkpoints/generator.ckpt')

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 64
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 4

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/4... Discriminator Loss: 2.2634... Generator Loss: 0.2027
Epoch 1/4... Discriminator Loss: 1.7997... Generator Loss: 0.3287
Epoch 1/4... Discriminator Loss: 1.5909... Generator Loss: 0.4161
Epoch 1/4... Discriminator Loss: 1.0862... Generator Loss: 0.7183
Epoch 1/4... Discriminator Loss: 0.8602... Generator Loss: 0.8826
Epoch 1/4... Discriminator Loss: 0.9453... Generator Loss: 0.7518
Epoch 1/4... Discriminator Loss: 1.0383... Generator Loss: 0.6631
Epoch 1/4... Discriminator Loss: 0.6143... Generator Loss: 1.2444
Epoch 1/4... Discriminator Loss: 0.7419... Generator Loss: 1.0538
Epoch 1/4... Discriminator Loss: 0.3646... Generator Loss: 1.6953

Epoch 1/4... Discriminator Loss: 0.7839... Generator Loss: 0.8982
Epoch 1/4... Discriminator Loss: 0.7520... Generator Loss: 1.0184
Epoch 1/4... Discriminator Loss: 0.6859... Generator Loss: 1.1878
Epoch 1/4... Discriminator Loss: 0.7980... Generator Loss: 1.1074
Epoch 1/4... Discriminator Loss: 1.3577... Generator Loss: 0.7905
Epoch 1/4... Discriminator Loss: 1.3006... Generator Loss: 0.7408
Epoch 1/4... Discriminator Loss: 1.2886... Generator Loss: 0.7081
Epoch 1/4... Discriminator Loss: 1.2474... Generator Loss: 0.7451
Epoch 1/4... Discriminator Loss: 1.2834... Generator Loss: 0.7232
Epoch 1/4... Discriminator Loss: 1.1897... Generator Loss: 0.7486

Epoch 1/4... Discriminator Loss: 1.1986... Generator Loss: 0.7848
Epoch 1/4... Discriminator Loss: 1.2812... Generator Loss: 0.7246
Epoch 1/4... Discriminator Loss: 1.3089... Generator Loss: 0.7161
Epoch 1/4... Discriminator Loss: 1.3271... Generator Loss: 0.7115
Epoch 1/4... Discriminator Loss: 1.2277... Generator Loss: 0.7458
Epoch 1/4... Discriminator Loss: 1.3186... Generator Loss: 0.7107
Epoch 1/4... Discriminator Loss: 1.2730... Generator Loss: 0.7266
Epoch 1/4... Discriminator Loss: 1.3153... Generator Loss: 0.7207
Epoch 1/4... Discriminator Loss: 1.3481... Generator Loss: 0.7140
Epoch 1/4... Discriminator Loss: 1.2972... Generator Loss: 0.7128

Epoch 1/4... Discriminator Loss: 1.3110... Generator Loss: 0.7172
Epoch 1/4... Discriminator Loss: 1.2680... Generator Loss: 0.7142
Epoch 1/4... Discriminator Loss: 1.2746... Generator Loss: 0.7265
Epoch 1/4... Discriminator Loss: 1.3043... Generator Loss: 0.6828
Epoch 1/4... Discriminator Loss: 1.2684... Generator Loss: 0.7492
Epoch 1/4... Discriminator Loss: 1.3308... Generator Loss: 0.6800
Epoch 1/4... Discriminator Loss: 1.3054... Generator Loss: 0.7089
Epoch 1/4... Discriminator Loss: 1.2557... Generator Loss: 0.7358
Epoch 1/4... Discriminator Loss: 1.3775... Generator Loss: 0.6893
Epoch 1/4... Discriminator Loss: 1.3075... Generator Loss: 0.6979

Epoch 1/4... Discriminator Loss: 1.2527... Generator Loss: 0.7404
Epoch 1/4... Discriminator Loss: 1.2787... Generator Loss: 0.7156
Epoch 1/4... Discriminator Loss: 1.3353... Generator Loss: 0.7121
Epoch 1/4... Discriminator Loss: 1.3077... Generator Loss: 0.7156
Epoch 1/4... Discriminator Loss: 1.1946... Generator Loss: 0.7725
Epoch 1/4... Discriminator Loss: 1.2701... Generator Loss: 0.7120
Epoch 1/4... Discriminator Loss: 1.3353... Generator Loss: 0.6548
Epoch 1/4... Discriminator Loss: 1.1874... Generator Loss: 0.7732
Epoch 1/4... Discriminator Loss: 1.3367... Generator Loss: 0.6940
Epoch 1/4... Discriminator Loss: 1.3259... Generator Loss: 0.6920

Epoch 1/4... Discriminator Loss: 1.3115... Generator Loss: 0.6945
Epoch 1/4... Discriminator Loss: 1.3250... Generator Loss: 0.7186
Epoch 1/4... Discriminator Loss: 1.2990... Generator Loss: 0.7080
Epoch 1/4... Discriminator Loss: 1.2839... Generator Loss: 0.7154
Epoch 1/4... Discriminator Loss: 1.2690... Generator Loss: 0.7208
Epoch 1/4... Discriminator Loss: 1.2912... Generator Loss: 0.7433
Epoch 1/4... Discriminator Loss: 1.3220... Generator Loss: 0.7231
Epoch 1/4... Discriminator Loss: 1.3419... Generator Loss: 0.6932
Epoch 1/4... Discriminator Loss: 1.2853... Generator Loss: 0.6968
Epoch 1/4... Discriminator Loss: 1.2817... Generator Loss: 0.7098

Epoch 1/4... Discriminator Loss: 1.3053... Generator Loss: 0.7198
Epoch 1/4... Discriminator Loss: 1.2983... Generator Loss: 0.7063
Epoch 1/4... Discriminator Loss: 1.2788... Generator Loss: 0.7306
Epoch 1/4... Discriminator Loss: 1.2596... Generator Loss: 0.7275
Epoch 1/4... Discriminator Loss: 1.3140... Generator Loss: 0.6882
Epoch 1/4... Discriminator Loss: 1.3251... Generator Loss: 0.7086
Epoch 1/4... Discriminator Loss: 1.3388... Generator Loss: 0.7051
Epoch 1/4... Discriminator Loss: 1.4108... Generator Loss: 0.6851
Epoch 1/4... Discriminator Loss: 1.3293... Generator Loss: 0.6585
Epoch 1/4... Discriminator Loss: 1.2888... Generator Loss: 0.7274

Epoch 1/4... Discriminator Loss: 1.2832... Generator Loss: 0.7025
Epoch 1/4... Discriminator Loss: 1.2475... Generator Loss: 0.7481
Epoch 1/4... Discriminator Loss: 1.2446... Generator Loss: 0.7549
Epoch 1/4... Discriminator Loss: 1.2222... Generator Loss: 0.7599
Epoch 1/4... Discriminator Loss: 1.3030... Generator Loss: 0.7125
Epoch 1/4... Discriminator Loss: 1.2500... Generator Loss: 0.7468
Epoch 1/4... Discriminator Loss: 1.2626... Generator Loss: 0.7160
Epoch 1/4... Discriminator Loss: 1.3138... Generator Loss: 0.7063
Epoch 1/4... Discriminator Loss: 1.2527... Generator Loss: 0.7565
Epoch 1/4... Discriminator Loss: 1.2802... Generator Loss: 0.7096

Epoch 1/4... Discriminator Loss: 1.2443... Generator Loss: 0.7396
Epoch 1/4... Discriminator Loss: 1.2410... Generator Loss: 0.7894
Epoch 1/4... Discriminator Loss: 1.1959... Generator Loss: 0.7691
Epoch 1/4... Discriminator Loss: 1.2020... Generator Loss: 0.7410
Epoch 1/4... Discriminator Loss: 1.2374... Generator Loss: 0.7460
Epoch 1/4... Discriminator Loss: 1.2502... Generator Loss: 0.7265
Epoch 1/4... Discriminator Loss: 1.3272... Generator Loss: 0.6906
Epoch 1/4... Discriminator Loss: 1.2162... Generator Loss: 0.7823
Epoch 1/4... Discriminator Loss: 1.2728... Generator Loss: 0.7084
Epoch 1/4... Discriminator Loss: 1.2507... Generator Loss: 0.7441

Epoch 1/4... Discriminator Loss: 1.2494... Generator Loss: 0.7574
Epoch 1/4... Discriminator Loss: 1.2911... Generator Loss: 0.7192
Epoch 1/4... Discriminator Loss: 1.2762... Generator Loss: 0.7576
Epoch 2/4... Discriminator Loss: 1.3469... Generator Loss: 0.6749
Epoch 2/4... Discriminator Loss: 1.1849... Generator Loss: 0.7686
Epoch 2/4... Discriminator Loss: 1.2610... Generator Loss: 0.7721
Epoch 2/4... Discriminator Loss: 1.2129... Generator Loss: 0.7837
Epoch 2/4... Discriminator Loss: 1.2740... Generator Loss: 0.7451
Epoch 2/4... Discriminator Loss: 1.3038... Generator Loss: 0.7514
Epoch 2/4... Discriminator Loss: 1.2576... Generator Loss: 0.7090

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 8

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/8... Discriminator Loss: 3.8353... Generator Loss: 0.0454
Epoch 1/8... Discriminator Loss: 2.8799... Generator Loss: 0.1106
Epoch 1/8... Discriminator Loss: 2.1028... Generator Loss: 0.2232
Epoch 1/8... Discriminator Loss: 2.2372... Generator Loss: 0.2059
Epoch 1/8... Discriminator Loss: 2.0793... Generator Loss: 0.2202
Epoch 1/8... Discriminator Loss: 1.8480... Generator Loss: 0.3013
Epoch 1/8... Discriminator Loss: 1.6398... Generator Loss: 0.3526
Epoch 1/8... Discriminator Loss: 1.4854... Generator Loss: 0.3980
Epoch 1/8... Discriminator Loss: 1.3248... Generator Loss: 0.4767
Epoch 1/8... Discriminator Loss: 1.0869... Generator Loss: 0.6144

Epoch 1/8... Discriminator Loss: 1.0395... Generator Loss: 0.6428
Epoch 1/8... Discriminator Loss: 0.7878... Generator Loss: 0.8642
Epoch 1/8... Discriminator Loss: 0.7263... Generator Loss: 0.8876
Epoch 1/8... Discriminator Loss: 0.6644... Generator Loss: 0.9989
Epoch 1/8... Discriminator Loss: 0.6685... Generator Loss: 1.0291
Epoch 1/8... Discriminator Loss: 0.6731... Generator Loss: 1.1495
Epoch 1/8... Discriminator Loss: 0.7541... Generator Loss: 0.9296
Epoch 1/8... Discriminator Loss: 0.7196... Generator Loss: 0.9772
Epoch 1/8... Discriminator Loss: 0.6290... Generator Loss: 1.2389
Epoch 1/8... Discriminator Loss: 0.6745... Generator Loss: 1.1084

Epoch 1/8... Discriminator Loss: 0.7126... Generator Loss: 1.1013
Epoch 1/8... Discriminator Loss: 0.8295... Generator Loss: 0.8181
Epoch 1/8... Discriminator Loss: 0.8056... Generator Loss: 1.0090
Epoch 1/8... Discriminator Loss: 0.5916... Generator Loss: 1.2791
Epoch 1/8... Discriminator Loss: 0.9607... Generator Loss: 0.7560
Epoch 1/8... Discriminator Loss: 0.6419... Generator Loss: 1.2697
Epoch 1/8... Discriminator Loss: 0.7757... Generator Loss: 1.0522
Epoch 1/8... Discriminator Loss: 0.5490... Generator Loss: 1.5908
Epoch 1/8... Discriminator Loss: 0.5294... Generator Loss: 1.4525
Epoch 1/8... Discriminator Loss: 0.4272... Generator Loss: 2.2999

Epoch 1/8... Discriminator Loss: 0.3904... Generator Loss: 2.0551
Epoch 1/8... Discriminator Loss: 0.3866... Generator Loss: 1.8779
Epoch 1/8... Discriminator Loss: 0.3527... Generator Loss: 2.5491
Epoch 1/8... Discriminator Loss: 0.4367... Generator Loss: 1.6511
Epoch 1/8... Discriminator Loss: 0.3532... Generator Loss: 2.0445
Epoch 1/8... Discriminator Loss: 0.3233... Generator Loss: 2.5832
Epoch 1/8... Discriminator Loss: 0.7816... Generator Loss: 0.9280
Epoch 1/8... Discriminator Loss: 0.2868... Generator Loss: 2.1513
Epoch 1/8... Discriminator Loss: 0.4230... Generator Loss: 2.3067
Epoch 1/8... Discriminator Loss: 0.3799... Generator Loss: 1.7940

Epoch 1/8... Discriminator Loss: 0.6467... Generator Loss: 1.0641
Epoch 1/8... Discriminator Loss: 0.4588... Generator Loss: 1.5223
Epoch 1/8... Discriminator Loss: 0.3660... Generator Loss: 1.6812
Epoch 1/8... Discriminator Loss: 0.5033... Generator Loss: 1.3288
Epoch 1/8... Discriminator Loss: 0.5599... Generator Loss: 1.4486
Epoch 1/8... Discriminator Loss: 0.5436... Generator Loss: 1.2270
Epoch 1/8... Discriminator Loss: 0.5231... Generator Loss: 1.8263
Epoch 1/8... Discriminator Loss: 0.1596... Generator Loss: 4.5789
Epoch 1/8... Discriminator Loss: 0.6058... Generator Loss: 1.1604
Epoch 1/8... Discriminator Loss: 0.3598... Generator Loss: 3.9623

Epoch 1/8... Discriminator Loss: 0.4058... Generator Loss: 2.2697
Epoch 1/8... Discriminator Loss: 0.2474... Generator Loss: 2.3316
Epoch 1/8... Discriminator Loss: 0.2172... Generator Loss: 3.2846
Epoch 1/8... Discriminator Loss: 0.3453... Generator Loss: 2.3377
Epoch 1/8... Discriminator Loss: 0.2268... Generator Loss: 2.5543
Epoch 1/8... Discriminator Loss: 0.2446... Generator Loss: 2.5617
Epoch 1/8... Discriminator Loss: 0.3479... Generator Loss: 1.7380
Epoch 1/8... Discriminator Loss: 0.1878... Generator Loss: 4.0042
Epoch 1/8... Discriminator Loss: 0.1517... Generator Loss: 3.9816
Epoch 1/8... Discriminator Loss: 0.3141... Generator Loss: 1.8880

Epoch 1/8... Discriminator Loss: 0.2012... Generator Loss: 2.5967
Epoch 1/8... Discriminator Loss: 1.2374... Generator Loss: 1.5005
Epoch 1/8... Discriminator Loss: 0.2344... Generator Loss: 2.3903
Epoch 1/8... Discriminator Loss: 0.2575... Generator Loss: 3.1788
Epoch 1/8... Discriminator Loss: 0.2990... Generator Loss: 2.3890
Epoch 1/8... Discriminator Loss: 0.1974... Generator Loss: 2.7040
Epoch 1/8... Discriminator Loss: 0.1828... Generator Loss: 2.9400
Epoch 1/8... Discriminator Loss: 0.2690... Generator Loss: 7.9752
Epoch 1/8... Discriminator Loss: 0.4578... Generator Loss: 1.9500
Epoch 1/8... Discriminator Loss: 0.2133... Generator Loss: 2.7794

Epoch 1/8... Discriminator Loss: 0.2186... Generator Loss: 2.9192
Epoch 1/8... Discriminator Loss: 0.4677... Generator Loss: 2.6099
Epoch 1/8... Discriminator Loss: 0.4066... Generator Loss: 1.9466
Epoch 1/8... Discriminator Loss: 0.2076... Generator Loss: 4.6814
Epoch 1/8... Discriminator Loss: 0.4023... Generator Loss: 1.6643
Epoch 1/8... Discriminator Loss: 0.2667... Generator Loss: 5.5084
Epoch 1/8... Discriminator Loss: 0.1635... Generator Loss: 3.9260
Epoch 1/8... Discriminator Loss: 0.2243... Generator Loss: 3.0980
Epoch 1/8... Discriminator Loss: 0.1626... Generator Loss: 3.1947
Epoch 1/8... Discriminator Loss: 0.2744... Generator Loss: 1.9937

Epoch 1/8... Discriminator Loss: 0.3438... Generator Loss: 2.1192
Epoch 1/8... Discriminator Loss: 0.2831... Generator Loss: 2.6358
Epoch 1/8... Discriminator Loss: 0.5006... Generator Loss: 1.4667
Epoch 1/8... Discriminator Loss: 0.4035... Generator Loss: 2.4688
Epoch 1/8... Discriminator Loss: 0.2852... Generator Loss: 2.4210
Epoch 1/8... Discriminator Loss: 0.3968... Generator Loss: 1.6942
Epoch 1/8... Discriminator Loss: 0.5134... Generator Loss: 1.4761
Epoch 1/8... Discriminator Loss: 0.3983... Generator Loss: 2.2121
Epoch 1/8... Discriminator Loss: 0.2093... Generator Loss: 3.5304
Epoch 1/8... Discriminator Loss: 0.4656... Generator Loss: 2.1838

Epoch 1/8... Discriminator Loss: 0.4117... Generator Loss: 1.8870
Epoch 1/8... Discriminator Loss: 0.4693... Generator Loss: 1.7517
Epoch 1/8... Discriminator Loss: 0.1848... Generator Loss: 4.4633
Epoch 1/8... Discriminator Loss: 0.2491... Generator Loss: 2.5479
Epoch 1/8... Discriminator Loss: 0.2697... Generator Loss: 4.7969
Epoch 1/8... Discriminator Loss: 0.2403... Generator Loss: 2.8573
Epoch 1/8... Discriminator Loss: 0.8792... Generator Loss: 0.7095
Epoch 1/8... Discriminator Loss: 1.0418... Generator Loss: 0.6738
Epoch 1/8... Discriminator Loss: 0.1484... Generator Loss: 4.6372
Epoch 1/8... Discriminator Loss: 0.2172... Generator Loss: 5.7121

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.